From Skills to Plans: Automatic Skill Discovery and Symbolic Interpretation for Compositional Tasks
Keywords: automatic skill discovery, symbolic interpretation, pixel-based controlling
TL;DR: We propose a novel pixel-based framework that combines entity-centric decision transformers with symbolic plannin to deal with long-horizon sequential task and real-world object manipulation.
Abstract: Deep Reinforcement Learning (DRL) has struggled with pixel-based controlling tasks that have numerous entities, long sequences, and logical dependencies. Methods using structured representations have shown promise in generalizing to different object entities in manipulation tasks. However, they lack the ability to segment and reuse basic skills. Neuro-symbolic RL excels in handling long sequential decomposable tasks yet heavily relies on expert-designed predicates. To address these challenges, we introduce a novel pixel-based framework that combines entity-centric decision transformers with symbolic planning. Our approach first automatically discovers and learns basic skills through experiences in simple environments without human intervention. Then, we employ a genetic algorithm to enhance these basic skills with symbolic interpretations. Therefore, we convert the complex controlling problem into a planning problem. Taking advantage of symbolic planning and entity-centric skills, our model is inherently interpretable and provides compositional generalizability. The results of the experiments show that our method demonstrates superior performance in long-horizon sequential tasks and real-world object manipulation.
Primary Area: neurosymbolic & hybrid AI systems (physics-informed, logic & formal reasoning, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 10729
Loading